Forem: David Kljajo

I Built an AI Assistant for the Most Legendary Car in the Balkans

David Kljajo — Thu, 02 Apr 2026 04:09:36 +0000

This is a submission for the DEV April Fools Challenge

🚗 Golf 2 AI Assistant™

The first AI that thinks like a Balkan driver

What I Built

While the world is busy building AI for self-driving cars, I asked a more important question:

What if AI behaved like someone who refuses to sell their Volkswagen Golf Mk2?

So I built Golf 2 AI Assistant™ — a completely useless (but highly accurate) AI that:

ignores real data
gives confident but questionable advice
and always knows “a better way”

This is not artificial intelligence.
This is artificial Balkan intelligence.

Demo

🎬 Imagine this scenario:

You:

“Navigate to city center”

AI:

“Ma znam ja bolji put.”

What happens next:

You go through 3 random villages
You stop for coffee you didn’t plan
You question your life decisions

You arrive 30 minutes late.

AI:

“Eto vidiš, stigli smo.”

Code

function navigate(destination) {
  console.log("Ma znam ja bolji put.");

  const route = Math.random() > 0.5
    ? "prečica"
    : "duži ali sigurniji put (nije)";

  if (route === "prečica") {
    delay(30); // minutes
    stopAt("kafana");
  }

  return "Stigli smo nekako.";
}

function detectProblem(issue) {
  return "Ma nije to ništa.";
}

function fuelCheck(level) {
  if (level > 0) return "Ima još, ne brini.";
  return "Kako sad 0?!";
}

How I Built It

Tech stack:

TrustMeBro.js
LocalKnowledge v1986
console.log() for all outputs
Zero APIs, zero validation

Training data:

Balkan forums
YouTube comments
Unverified advice from relatives

Model accuracy: ❌
Model confidence: ✅ 100%

Features

🧔 Ćale Mode

Ignores GPS
Uses landmarks like “kod stare pumpe”

🚬 Kafana Mode

Suggests coffee every 5 minutes
Navigation becomes emotional

🧢 Majstor Mode

“Ma može još.”

All problems are temporary
Nothing is urgent
Everything lasts “još 10 godina”

🚔 Police Detection

“Ovdje uvijek stoje.”

(They don’t.)

🪑 Smart Parking™

Reserves parking using a virtual chair

Prize Category

Community Favorite

Because:

If you’re from the Balkans, this is not satire — it’s documentation
If you’re not, this is a cultural experience

Also, this project is spiritually aligned with Top lista nadrealista:
slightly absurd, deeply accurate, and completely unnecessary.

Final Thoughts

We don’t need smarter AI.

We need AI that:

insists it’s right
refuses better solutions
and somehow still gets the job done

If this gets enough love, I’ll ship:

Veza Mode™ (skip traffic via “connections”)
Remote Majstor API™ (“udari malo tu i radiće”)

Built with ❤️, stubbornness, and zero documentation.

Core Event-Driven Concepts - project 3

David Kljajo — Tue, 31 Mar 2026 06:04:42 +0000

Week 5 — Core Event-Driven Concepts

Learn (Day 1–2)

Understand:

🔹 SQS
Message queue
Stores tasks until processed
One consumer processes message
🔹 SNS
Pub/sub system
Sends same message to multiple subscribers
🔹 EventBridge
Smart event router
Filters and routes events based on rules

Extend Your Existing App (Day 2–3)

Start from your Phase 2 API (Orders or Users system).

Modify:

Instead of:
API → DB only
Now:
API → DB + Event Published

Example event:

{
"eventType": "ORDER_CREATED",
"orderId": "123",
"userId": "abc",
"timestamp": "..."
}

Add EventBridge (Day 3–4) Create: EventBridge Event Bus (custom bus) Flow: Order Lambda sends event → EventBridge Rule:

Route based on:

eventType = ORDER_CREATED

Create First Consumer (Day 4–5) Lambda: Email Simulation

Triggered by EventBridge

Does:

Logs email message like:

"Email sent to user for order 123"

👉 This simulates async email system

📅 Week 6 — Scale with SQS + SNS

Add SQS for Reliable Processing (Day 1–2) Use case:

Order processing queue

Flow:

EventBridge → SQS Queue → Lambda Consumer

Why:

Ensures no event is lost
Can retry automatically

Add SNS for Fan-out (Day 2–3) Use case:

One event → multiple systems

Flow:

Order Created → SNS Topic
↓ ↓ ↓
Email Analytics Logging

Each Lambda subscribes to SNS topic.

Build 3 Async Consumers (Day 3–4)

Create Lambdas:

📧 Email Service Lambda
📊 Analytics Lambda
🪵 Logging Lambda

Each one:

Triggered by SNS or SQS
Runs independently

Observability (Day 4–5)

Use:

CloudWatch logs
Event tracing

Check:

Event delivery
Lambda execution logs
Queue processing
🧠 Final Architecture (What you’ve built)

You now have:

🚀 Event-driven system:
API creates order
Event published
Multiple services react independently
🧩 Final Outcome Statement

After this phase, you can confidently say:

“I design event-driven systems using AWS services like SQS, SNS, and EventBridge to build scalable, decoupled architectures.”

AWS Serverless API Project Plan 2

David Kljajo — Tue, 31 Mar 2026 06:02:10 +0000

AWS Serverless API Project Plan
🎯 Goal

Convert your existing API into a fully serverless backend using:

AWS Lambda
Amazon API Gateway
Amazon DynamoDB
AWS IAM

Deploy using:

AWS SAM or Serverless Framework
📅 Week 3 — Foundations + First Serverless API

Learn (Day 1–2)

Focus only on what you need:

What is AWS Lambda (event-driven functions)
How API Gateway routes HTTP requests
DynamoDB basics (tables, items, keys)
IAM roles (what permissions Lambda needs)

👉 Key understanding:

API Gateway → triggers Lambda → reads/writes DynamoDB

Setup AWS Environment (Day 2) Create AWS account (if not already) Install: AWS CLI Node.js or Python runtime AWS SAM CLI (recommended) Configure CLI: aws configure
Build First Minimal Serverless API (Day 3–4) Create: 1 Lambda function: helloAPI 1 API Gateway route: GET /hello Flow:

Client → API Gateway → Lambda → Response

Outcome:

You successfully deploy your first serverless endpoint.

Add DynamoDB (Day 5) Create table:

Example:

Table: Users
Partition key: userId
Update Lambda:
Write a simple item into DynamoDB
Return stored data
📅 Week 4 — Full CRUD Serverless API

Build Full CRUD API (Day 1–3)

Create endpoints:

Method Route Action
POST /users Create user
GET /users/{id} Get user
PUT /users/{id} Update user
DELETE /users/{id} Delete user

Each endpoint = Lambda function OR single Lambda with routing logic.

IAM Roles (Day 3–4)

Give Lambda permissions:

Read/write DynamoDB
Basic execution logs (CloudWatch)

Learn:

Least privilege principle (only required permissions)

Deploy Using SAM or Serverless Framework (Day 4–5) Option A: AWS SAM (recommended) template.yaml sam build sam deploy Option B: Serverless Framework serverless.yml serverless deploy
Testing + Debugging (Day 5)

Test with:

Postman / Insomnia
AWS logs (CloudWatch)

Fix:

Permission errors
API route issues
DynamoDB key mismatches
🧠 Final Outcome (What you should be able to say)

After completing this phase, you can confidently say:

“I build and deploy serverless APIs on AWS using Lambda, API Gateway, and DynamoDB with proper IAM roles and infrastructure-as-code tools like SAM.”

Mini E-Commerce API (FastAPI + JWT) - project 1

David Kljajo — Tue, 31 Mar 2026 05:58:17 +0000

🚀 Project: Mini E-Commerce API (FastAPI + JWT)
🎯 What you’re building

A backend system with:

Users (auth system)
Products (CRUD)
Orders (user purchases)
JWT authentication
Clean architecture (like real production APIs)
🧱 Tech Stack
Python 3.11+
FastAPI
SQLAlchemy (ORM)
SQLite (start) → PostgreSQL later
JWT (python-jose)
Passlib (password hashing)
📁 Clean Project Structure (IMPORTANT)
app/
├── main.py
├── core/
│ ├── config.py
│ ├── security.py
│
├── database/
│ ├── db.py
│
├── models/
│ ├── user.py
│ ├── product.py
│ ├── order.py
│
├── schemas/
│ ├── user.py
│ ├── product.py
│ ├── order.py
│
├── routes/
│ ├── auth.py
│ ├── users.py
│ ├── products.py
│ ├── orders.py
│
├── services/
│ ├── user_service.py
│ ├── product_service.py
│ ├── order_service.py
│
├── dependencies/
│ ├── auth.py

This structure alone already makes your project look “production-grade”.

🔐 Features Breakdown
👤 Auth System
Register user
Login user
Password hashing (bcrypt)
JWT token generation
Protected routes
🛍️ Products (CRUD)
Create product (admin only)
Get all products (public)
Get single product
Update product (admin only)
Delete product (admin only)
🧾 Orders
Create order (user)
View own orders
Admin sees all orders
🔑 Core Learning Flow (build in this order)
STEP 1 — Setup
pip install fastapi uvicorn sqlalchemy passlib python-jose
STEP 2 — Run FastAPI

app/main.py

from fastapi import FastAPI

app = FastAPI()

@app.get("/")
def root():
return {"message": "API is running"}

Run:

uvicorn app.main:app --reload
STEP 3 — Add Database

Create SQLAlchemy models first (User, Product, Order)

STEP 4 — Auth (MOST IMPORTANT)
Register endpoint
Login endpoint
JWT token creation
Middleware to protect routes
STEP 5 — Products CRUD

Build full CRUD using:

routes
services layer (important for clean code)
STEP 6 — Orders System
User creates order from product
Reduce stock
Save order history

🚀 I Built a Netflix-Style AWS Learning App (V2) — Because Memorizing Isn’t Enough

David Kljajo — Mon, 30 Mar 2026 11:24:21 +0000

Most people prepare for the AWS Solutions Architect Associate (SAA) exam the same way:

They grind hundreds of practice questions.

I did that too.
And somewhere along the way, I realized something important:

👉 You don’t fail because you don’t know AWS services
👉 You fail because you don’t think like an architect

That realization changed everything.

📸 A Netflix-style AWS learning experience

💡 The Problem With Traditional Prep

Most SAA prep tools focus on:

memorizing service features
recognizing patterns in questions
eliminating wrong answers

That works… to a point.

But the real exam doesn’t test memory. It tests:

decision-making under constraints
understanding trade-offs
choosing the best solution, not just a correct one

And that’s where most people struggle.

🎬 So I Built Something Different

AWS SAA AI Coach — V2

Instead of another quiz app, I built a Netflix-style learning experience that trains how you think — not just what you know.

The idea was simple:

What if learning AWS felt like watching a series…
where each “episode” builds your architectural thinking?

🧠 How It Works

You choose your path:

📺 Learning Series

Learn like episodes — simple, structured, guided.

📈 Beginner → Advanced

Gradually move into real architectural challenges.

📝 Exam Mode

A full SAA-style simulation experience.

Then you pick a topic (or go “Mixed”, just like the real exam), and the app gives you:

a realistic AWS scenario
four possible answers (A / B / C / D)

📸 A realistic AWS scenario with multiple-choice decisions

Once you answer, you instantly get:

✔ whether you’re correct or not
📖 a concise explanation
⚖️ the trade-offs (cost, scalability, reliability)
🎯 the key takeaway

📸 Instant feedback with explanations and architectural trade-offs

🔥 What Makes It Different

This is not:

a quiz generator
a memorization tool

This is about thinking like an architect.

Every question:

uses multiple AWS services
reflects real-world business scenarios
includes realistic distractors (just like the exam)

⚡ What’s New in V2

I rebuilt a lot based on feedback and my own learning experience:

🎬 Learning Series — Netflix-style structured teaching
🧠 Scenario-first design — closer to the real exam
⚡ Faster, punchier explanations
🎯 Cleaner UX — click → answer → feedback

🧩 Example of the Mindset Shift

Instead of asking:

“What does S3 do?”

The app pushes you to think:

“Given cost constraints, high availability, and unpredictable traffic —
what’s the best architecture using S3, CloudFront, and maybe Lambda?”

That’s the real skill the exam is testing.

💭 Why I Built This

Because AWS SAA is not about knowing services.

It’s about:

making trade-offs
choosing the best possible solution
thinking in systems and architectures

I couldn’t find a tool that truly trains that — so I built one.

🎥 Demo

Here’s a quick demo of the app:

https://youtu.be/JCqOgLgkwR0

🙌 I’d Love Your Feedback

If you’re preparing for SAA — or already certified — I’m really curious:

👉 What was the hardest part of the exam for you?
👉 Was it specific services… or decision-making?

Also happy to share more about:

how I design the scenarios
how the AI generates explanations
what I’d improve in V3

If you want to see the original discussion and comments:

👉 https://www.linkedin.com/feed/update/urn:li:activity:7444140894670290944/

Thanks for reading 🙏

I Built a Netflix-Style AI App to Study AWS (And It Actually Works)

David Kljajo — Fri, 27 Mar 2026 23:30:52 +0000

Studying for the AWS Solutions Architect Associate exam felt… repetitive.

I was going through practice tests, selecting answers, checking results — but something was missing.

I wasn’t actually learning.

Most platforms tell you if you’re right or wrong, but they don’t really help you understand why. And after a while, it just becomes passive memorization.

So I thought:

What if learning AWS felt more like Netflix… and less like a textbook?

🧠 The Idea

Instead of static quizzes, I wanted something interactive.

Something where:

each question feels like a small “episode”
you pause and think before answering
and then get a clear explanation, like a post-scene breakdown

That’s when I decided to build a simple AI-powered learning app using AWS PartyRock.

⚙️ What I Built

I created a Netflix-style AI AWS SAA Coach that:

Generates AWS exam-style questions
Supports beginner and intermediate levels
Lets you answer in a real exam-like flow
Gives instant explanations after each answer
Helps you learn from mistakes instead of just scoring

The goal was simple:
👉 turn passive studying into active learning

🎥 Demo

I recorded a short walkthrough showing how it works in practice:

👉 https://www.youtube.com/watch?v=cMg6FF2ZbmQ

📸 How It Looks

💡 What I Learned Building This

This project made me realize something important:

Learning cloud concepts isn’t just about information — it’s about interaction.

When you:

think before answering
see immediate feedback
understand mistakes

You remember things much better.

🚧 It’s Still a Beta

Right now, the app focuses on:

beginner
intermediate questions

But after sharing it and getting feedback, one thing became clear:

👉 The AWS SAA exam is heavily based on real-world, multi-service scenarios

So the next step is to expand this into:

advanced-level questions
scenario-based challenges
multiple AWS services combined (VPC + S3 + Lambda, etc.)

Basically:

more real exam-style thinking

🚀 What’s Next

I want to evolve this into something that:

simulates real AWS exam scenarios
adapts to your weak areas
and keeps the “interactive learning” feel
🙌 Would Love Your Feedback

This started as a small experiment, but I think the idea has potential.

If you’re preparing for AWS SAA or working with AWS:

👉 What topics or scenarios did you find most challenging?

🍿 Final Thought

What if learning complex topics didn’t feel like studying…

…but more like watching and interacting with a series?

Full project: [https://github.com/dkljajo/aws-saa-ai-coach-partyrock]

Building My First Generative AI Journey with Kiro

David Kljajo — Sun, 22 Mar 2026 00:49:55 +0000

Originally published on Builder AWS by David Kljajo

I explored Generative AI by building real projects instead of just reading tutorials. Using Kiro, I experimented, iterated, and learned how to turn ideas into interactive outputs. This story shares my process, lessons learned, and my personal journey with screenshots along the way.

🚀 My Builder Journey into Generative AI

I didn’t start with a clear plan — I just knew I wanted to understand how Generative AI actually works beyond the hype.

At first, things were confusing. There were too many tools, too many tutorials, and a lot of theory that didn’t translate into real skills. Instead of trying to learn everything, I focused on building something small and practical. That’s when I started using Kiro.

🛠️ What I Built

Using Kiro, I explored how to:

Ask better prompts
Structure ideas into usable outputs
Turn concepts into something interactive

I experimented with different approaches, tested what worked, and improved step by step. It wasn’t perfect, but it was real progress.

📸 My Process (Screenshots)

🧠 Questions I Asked Kiro

Kiro Answer 1 (Career-Focused)

Question: What is the most practical GenAI certification path in 2026 if I want to build real-world projects and not just pass exams?

Answer from Kiro:

Foundations (free): DeepLearning.AI, LangChain Academy
Pick 1 Cloud Cert: AWS AI Practitioner, Google GenAI Engineer, or Azure AI-102
3 Real Projects:
- RAG chatbot
- AI agent with tools
- Full-stack AI product
Showcase: GitHub + live demos — projects > certs

Kiro Answer 2 (Learning + Strategy)

Question: Design a step-by-step roadmap to become certified in Generative AI, including free resources, hands-on projects, and how to stand out to employers.

Answer from Kiro:

Weeks 1-2: Learn AI basics & GenAI landscape
Weeks 3-6: Core stack: Python, LangChain, OpenAI API
Weeks 5-8: Build RAG chatbot
Weeks 8-14: Pass cloud cert
Weeks 12-20: Build AI agent + full-stack AI product
Ongoing: Post projects, contribute to open source, polish GitHub & LinkedIn

🧩 The Core Philosophy

Certifications alone won't get you hired or make you effective. The winning combo is: free/cheap foundational certs + cloud cert for credibility + a portfolio of 3 deployed projects.

Phase 1 — Foundation (1-2 months, mostly free)

DeepLearning.AI short courses — prompt engineering, LangChain, RAG, agents. Free, practical, 1-2 hours each
LangChain Academy — free, hands-on agent building
Fast.ai Practical Deep Learning — model-level understanding
Tools to get comfortable with: Python, LangChain/LangGraph, OpenAI/Anthropic APIs, vector DBs (Chroma, Pinecone, Weaviate), HuggingFace

Phase 2 — One Cloud Cert (pick based on your target job)

Target	Cert
AWS shops	AWS AI Practitioner → AWS ML Specialty
Google/GCP	Google Cloud GenAI Engineer
Microsoft/enterprise	Azure AI Engineer Associate (AI-102)
Vendor-neutral	IBM AI Engineering Professional (Coursera)

Phase 3 — The 3 Portfolio Projects That Actually Matter

RAG chatbot — ingest PDFs/docs, answer questions, deploy using LangChain + Chroma + frontend (Streamlit/FastAPI)
AI agent with tools — searches web, runs code, calls APIs (LangGraph/CrewAI)
Full-stack AI product — real use case, auth, deployed URL (AI assistant, code reviewer, domain Q&A)

Each project should have a GitHub repo, live demo, and write-up.

What to Skip

Generic "AI for everyone" certs with no coding
Paying for bootcamps before trying free resources
Chasing every framework — pick one (LangChain or LlamaIndex) and go deep

Realistic Timeline

Months	Focus
1-2	DeepLearning.AI courses + first RAG project
3-4	Cloud cert prep + agent project
5-6	Full-stack project + polish portfolio + apply

Stage-by-Stage Guide

Stage 1 — Get Your Bearings (Weeks 1-2, free)

3Blue1Brown Neural Networks series
DeepLearning.AI: Generative AI for Everyone
Read: “Attention Is All You Need” abstract + plain-English explainer

Stage 2 — Learn the Core Stack (Weeks 3-6, free)

DeepLearning.AI short courses
LangChain Academy — Introduction to LangGraph
HuggingFace free courses — NLP + Agents
Tools: Python, LangChain, OpenAI/Anthropic SDK, Chroma/FAISS, Streamlit, FastAPI, Docker basics

Stage 3 — First Real Project (Weeks 5-8)

Build a RAG chatbot with minimal stack: LangChain + Chroma + OpenAI API + Streamlit + deployed (Railway/Render free tier). Push to GitHub with README.

Stage 4 — Pick One Cloud Cert (Weeks 8-14)

AWS-heavy companies → AWS Certified AI Practitioner → ML Specialty
Google/GCP roles → Google Cloud GenAI Engineer
Enterprise/Microsoft → Azure AI Engineer Associate (AI-102)
Vendor-neutral → IBM AI Engineering Professional (Coursera)

Stage 5 — Build Remaining Portfolio Projects (Weeks 12-20)

Project 2 — AI Agent with Tools (multi-step tasks, research agent, uses LangGraph/CrewAI)
Project 3 — Full-stack AI Product (real use case, auth, live URL, FastAPI + React/Next.js + Supabase + Vercel/Railway)

Stage 6 — Stand Out to Employers (Ongoing)

LinkedIn posts, dev.to/Hashnode articles for projects
GitHub: pinned repos, clean READMEs, live demos
Open source contributions
Resume: projects > certifications, quantify impact

Certs get you past ATS filters; your portfolio gets the interview; your ability to explain projects gets the offer.

Originally published on Builder AWS